A decision is not one event. It is a relay — someone decides, someone writes it down, someone builds it, someone ships it — and the baton gets dropped in five places. None of the five is exotic. All five are invisible from inside the org, which is why they survive so long.

I have been the person who owns the date since 1997, in launch teams at Microsoft, in a fifty-person engineering org at Kabam, across five departments and two business units at Verizon, and inside a hundred-and-twenty-person program at Kinoo. Different companies, different decades, different industries. The breaks were in the same five places every time.

Five places a decision dies A hand-drawn line runs from decided toward shipped and breaks five times: no owner, no definition, no decision, no communication, no capacity. At the fifth the ink runs out, and shipped is left unreached. decided shipped no owner no definition no decision no communication no capacity decided in a meeting, and it isn't in anyone's queue. "done" means something different to everyone in the room. the thing nobody wanted to raise, so it was never called. handoffs between departments, decisions between layers of management, and the tools that were supposed to capture both. the skill isn't on staff yet and nobody has hired for it. Five places a decision dies A hand-drawn line runs from decided toward shipped and breaks five times: no owner, no definition, no decision, no communication, no capacity. At the fifth the ink runs out, and shipped is left unreached. decided shipped no owner no definition no decision no communication no capacity decided in a meeting, and it isn't in anyone's queue. "done" means something different to everyone in the room. the thing nobody wanted to raise, so it was never called. handoffs between departments, decisions between layers of management, and the tools that were supposed to capture both. the skill isn't on staff yet and nobody has hired for it.
Every product I've been brought into breaks in these five places. Naming yours takes three weeks. Fixing them is a different engagement, and one I'll take on — with people I've worked with, briefed from the read.

What follows is each break, what it looks like on the ground, and what it costs. If you only read one section, read the fifth — it is the one that gets recorded in the retro as "we ran out of time".

Break oneNo owner

Decided in a meeting, and it isn't in anyone's queue.

The decision feels made because everyone senior was in the room and everyone nodded. What did not happen is the step where a decision becomes somebody's next piece of work — written into a list that person actually opens on Monday. A decision that exists only as a memory in a room is not a decision; it is a rumour with good attendance.

What it costs: the item waits, silently, until someone asks about it three weeks later. Then it gets re-litigated by people who were not in the room, because there is nothing to point at.

Break twoNo definition

"Done" means something different to everyone in the room.

Ask five people what finished looks like and you get five answers that are all defensible. The build then succeeds against a definition nobody who cared about the outcome ever read. This is the most expensive break because it fails late and looks like success on the way there — the burn-down is clean, the demo works, the thing is wrong.

What it costs: rework, or worse, shipping the wrong thing on time and discovering it in the quarter after.

Break threeNo decision

The thing nobody wanted to raise, so it was never called.

Every plan contains a trade-off somebody declined to make out loud. Two things cannot both be first, so the org does one and a half of them and calls the half an MVP. The trade-off does not disappear when it is not named; it gets decided by whoever is most tired, or by whoever ships last.

What it costs: a hidden assumption carried through the plan that fails in month four, when it is too late to choose differently and too early to blame anyone.

Break fourNo communication

Handoffs between departments, decisions between layers of management, and the tools that were supposed to capture both.

A decision travels through at least three vocabularies — the one it was made in, the one it was written in, and the one it was built in. Each translation loses a constraint. The ticket says build the button; the decision was build the button only for accounts in arrears; the constraint lived in a conversation nobody recorded.

What it costs: the same conversation, three times, with the second version differing from the first. Then a rebuild nobody budgeted for.

Break fiveNo capacity

The skill isn't on staff yet, and nobody has hired for it.

The plan assumes a person who does not exist. Not a person who is busy — a person whose job description has not been written. Sometimes it is a discipline the company has never had, sometimes it is a role that was cut in the last round and quietly re-assumed by the plan.

What it costs: the schedule risk everyone talks about converts into a scope risk nobody planned for, usually in the week the date stops being negotiable.

Why five, and not oneThe breaks compound

Any one of these is survivable. The reason they matter as a set is that each one makes the next more likely. A decision with no owner never gets a definition. A definition nobody agreed never gets communicated. A decision nobody communicated gets re-made by whoever has capacity — which is rarely the person with the skill.

By the fifth break, the org is not late because it built slowly. It is late because it decided the same thing four times and built a fraction of it once. That is why the fix is never "more process". Process is what you add when you have given up on the first two breaks.

Fixing these is not a process problem. It is an attention problem — someone has to look at where the decision actually goes.— Mike Jones

A test you can run this weekFive questions, no consultant required

Take the single most important thing your org is building right now and answer these five questions honestly. You will know within a day which break is yours.

  1. Name the decision and the person whose queue it is in. One name. If two, you have no owner.
  2. Write the sentence that means done. Then ask three people to say it back without reading it.
  3. Name the trade-off you deferred. What did you implicitly choose by not choosing?
  4. Point at where the decision is written down. Then point at who reads that page, and when.
  5. Name the person on the plan who does not exist yet. Then name the date the requisition opens.

If the answers take more than an hour to assemble, that is the answer. The work is not the fixing; it is the finding, and finding takes an outsider who is allowed to ask twice.

Where this fitsThree weeks, one document

My read exists because naming these five in someone else's org takes about three weeks. Interviews — you, your leads, three of your customers. The paper trail — the decision log, two quarters of roadmap, the actual work queue rather than the one in the tool. Then one written document saying which of the five is yours, what it has cost you this year, and what I would do about each.

It is a fixed fee, one client at a time, and the numbers are published rather than negotiated. If the read is wrong, you will know in the first ten minutes of the argument that comes with it.